Twisted cotangent bundle of hyperkähler manifolds (with an appendix by Simone Diverio) (Q2234120)

Let \((X,H)\) be a polarized manifold and denote by \(\pi:\mathbb{P}(\Omega_{X}) \to X\) the projectivization of the cotangent bundle, and by \(\zeta :=c_1(\mathscr{O}_{\Omega_X}(1))\) the tautological class on \(\mathbb{P}(\Omega_X)\). The twisted cotangent bundle \(\Omega_X \otimes H\) of the polarized manifold \((X,H)\) is pseudoeffective (resp. nef) if, by definition, \(\zeta + \pi^{*}H\) has this property. The main achievement of this paper is a sufficient condition for the pseudoeffectivity of the twisted cotangent bundle \(\Omega_X \otimes H\) of an Hyperkähler manifold, in terms of the Beauville-Bogomolov-Fujiki form \(q(H)\). An analogue result for \(K3\) surfaces is proven by Gounelas and Ottem [\textit{F. Gounelas} and \textit{J. C. Ottem}, Épijournal de Géom. Algébr., EPIGA 4, Article No. 8, 16 p. (2020; Zbl 1451.14119)] and this implies that if \((X, H)\) is a primitively polarized \(K3\) surface and \(q(H) \geq 8\) then \(\zeta+\pi^{*}(H)\) is pseudoeffective. In view of this result the authors state a conjecture: for a fixed even natural number \(2n\), there exists only finitely many deformation families of polarized Hyperkähler manifolds \((X,H)\), with \(\dim X = 2n\) and \(H\) an ample Cartier divisor on \(X\), such that \(\zeta+\pi^{*}(H)\) is not pseudoeffective. The main result of the paper is Theorem 1.3 where the authors prove that if \(X\) is an Hyperkähler manifold, there exist a constant \(C_X \geq 0\), depending only on the deformation family of \(X\), such that if \(X\) is very general in its deformation space and \(\omega_X\) is big and nef 1-1 class on \(X\), then \(q(\omega_X) \geq C_X\) if and only if \(\zeta+\pi^{*}\omega_X\) is nef. Moreover, in the same theorem, using a degeneration argument by Simone Diverio contained in the Appendix, they show that if \(q(\omega_X) \geq C_X\), then \(\zeta+\pi^{*}\omega_X\) is pseudoeffective. Corollary 1.4 is a first evidence of the previous conjecture for polarized Hyperkähler manifolds. Section 3 is devoted to the proof of Theorem 1.3. Section 4 concerns the proof of Theorem 1.5 which is a more explicit version of Theorem 1.3. Namely the authors describe how to compute the constant \(C_X\). Consider \(X\) an Hyperkähler manifold of dimension \(2n\), and assume that a very general deformation of \(X\) does not contain any proper subvarieties. If \(\omega_X\) is a Kähler calss on \(X\) and if \((\zeta+\lambda\pi^{*}\omega_X)^{4n-1} >0\) for all \(\lambda >1\), then \(\zeta+\pi^{*}\omega_X\) is pseudoeffective. Moreover they prove that if \(X\) is very general in its deformation space, then \(\zeta+\pi^{*}\omega_X\) is nef if and only if \((\zeta+\lambda\pi^{*}\omega_X)^{4n-1}>0\) for all \(\lambda>1\). The proof is based on the fact that the constant \(C_X\) in Theorem 1.3 is the largest real root of the polynomial \(p_X(t)\) defined by \(p_X(q(\omega))=(\zeta + \pi^{*}\omega)^{4n-1}\), and observing that the condition \(q(\omega_X) \geq C_X\) is equivalent to \((\zeta+\lambda\pi^{*}\omega_X)^{4n-1}>0\) for all \(\lambda>1\). The authors recall a result of Verbitsky that states that a very general deformation of the Hilbert scheme of length \(n\) on a \(K3\) surface does not contain any proper subvarieties. Moreover in Section 4 the authors associate to an Hyperkähler manifold \(X\) of dimension \(2n\), and a big and nef class \(\omega_X\) on \(X\), a positivity threshold \(\gamma_p(\omega_X) := \text{inf}\{\lambda_{0}\in \mathbb{R} \ | \ (\zeta + \lambda\pi^{*}\omega_X)^{4n-1} >0 \ \forall \ \lambda >\lambda_0\}.\) In Section 5 the constant \(C_X\) is computed for Hyperkähelr manifolds of type \(K3^{[2]}\) and the positivity threshold is explicited for Hyperkähler fourfolds. In Section 6 the positivity threshold is computed for Hyperkähler manifolds of type \(K3^{[3]}\) using a proposition communicated to the authors by Samule Boissière.